TMS320F2809: TMS320F2809 Analog Input acting as an output

Glenn,

If an analog pin is unloaded/open and is sampled there will be a feedback voltage from the internal S/H cap that will present itself on the associated ADCIN pin. 

However, this is usually along the lines of 1.xxV, and certainly can't(shouldn't?) exceed the supply rail of the device, VDDA/VDDIO(3.3V nom).

So, while the pheonomena of a voltage on the ADCIN pin is not unheard of, it certainly shouldn't be what you are seeing.  Can you comment on the state of the following:

1)Supply rails VDDIO/VDDA when you observe the issue

2)State of the other ADCIN pins, meaning is one of the other pins over-driven >>VDDA rail?  While this in itself is a violation of the DS, if we get over 0.3V above the VDDA rail we could reverse bias the internal S/H switch and potentially give a path to the other pins.

3)If tied to a specific sampling configuration, please list that as well.  Again if there is an overdriven pin, and it is sampled, it could reverse bias the same mux/switch mentioned in #2.

Best,

Matthew

Mathew,

Thanks for your advice.  It turns out I found out one of the options you suggested was true.

"2)State of the other ADCIN pins, meaning is one of the other pins over-driven >>VDDA rail?  While this in itself is a violation of the DS, if we get over 0.3V above the VDDA rail we could reverse bias the internal S/H switch and potentially give a path to the other pins."

I had another analog input which is driven by differential amplifier that is amplifying the differential signal supplied from another analog mux.  Under normal conditions this circuit never exceeds 3VDC, but occasionally we were disabling the mux and the signals fed to the differential amplifier are not controlled and since it has a supply voltage of 9.5VDC we can exceed recommended limit for analog inputs.  I measured  a 4.2VDC pulse on this analog input in one instance that occurred at the same time as the pulse I was seeing on the other analog pulse..  I also verified that the enable line was going high on the analog mux at the same time periods.  Typically this voltage doesn't last more than 200ms. 

My question is do I have to worry about possible damage to the adc circuit at a certain voltage or is this unlikely?  We are not aware of damage on this circuit yet.

We currently looking at leaving the analog mux enabled at all times to stop this from occurring.

Thanks,

Glenn Steffen

Glenn,

I'm glad we were able to track this down, power states across different domains/logic is always challenging to find every corner case.

In terms of damage, we need to determine the max current the amplifier can source.  Looking at the F2809 DS electrical specs, the clamp current of the ESD diodes on the device is +/-20mA total and +/-2mA per pin.  Meaning once we get ~0.3V above VDDA the device is going to start to turn these on in an attempt  to keep the voltage in the VDDA range.

If we exceed the 2mA(or if multiple pins >20mA) that is when we risk some damage to the device, specifically local to that pin/diode eventually it will destroy the local metal system typically resulting in either a pin open or pin short.  Even if we don't see immediate damage, over time this will degrade the circuit.

So, if we can keep the current within this limit, we should be OK.  Another option would be to place an Schottky diode externally to limit the voltage the pin will see if the events you mention take place(if it is too difficult to get the mux to control the situation).

Best,
Matthew